Why no new accelerators?

[rkauer]

Yeah. I know that running a latest mask 060 @100MHz is feasible (as the defaced "A" guys made with the CT060). The main problem is the memory + glue logic running even @ half that speed (using the 040 position on Apollos).

40ns SIMM sticks are not common at all...

 

[roy_bates]

Yeah. I know that running a latest mask 060 @100MHz is feasible (as the defaced "A" guys made with the CT060). The main problem is the memory + glue logic running even @ half that speed (using the 040 position on Apollos).

40ns SIMM sticks are not common at all...

i use 60ns ram on the apollo,the rest of the card apart from the 74F74 and 060 runs at 50mhz,blizzards are different as you cant half the bus,so yes your right 40ns would be pretty hard to get or should that be impossible.

i was going to write a wall of text about the CT60 but its not really worth it,i will say this though at least the guy dident give up just becuase it was connected to a 16 bit bus.


as for the peaple that say fpga is the way to go................i still beleive it will be years before its seen on the amiga if at all.(as an 060)

 

[Mr B]

as for the peaple that say fpga is the way to go................i still beleive it will be years before its seen on the amiga if at all.(as an 060)

I'm sadly thinking your right. Well, about the .060 part. In a couple of generations old FPGA stock will be powerfull enough to run 000-030 softcores at "real" speeds. When it gets cheap enough (when people no longer want to have them as storage heaters) they have to end up somewhere, and if we are still around, it makes sense to use them to replace hard to find CPU's with cheaper if yet more complex counterparts. I guess only time will tell, but thats what i think. And quite frankly, i don't really see a huge market for 68060 based parts softcore or other-ways, where as a 68030 will keep on selling for as long as there is Amigas without one... You don't really need more when all you do is WHDLoad, and (i might be biased) that seams to be what most people do. Amiga OS4 does enjoy a faster CPU still, but thats a more limited user-group. You guys are more exlusive, while the general public wants the retro experience, to relive some youth.
B!

 

[alexh]

It will be years before the price of an FPGA capable of taking a the full 060 design and running at plus 100MHz will be commercially attractive over the (now €80) MC68060RC50

The problem is not only the price of the FPGA's but also the availability of the design (i.e. it is closed source) and the lack of people interested enough to reverse engineer and write HDL Synthesiseable CPU cores.

It's what six years since MiniMig first came out?? Only 3-4 engineers have ever said they've been actively working on 680x0 HDL designs. And only Tobias Gubener's TG68 has been released. Nothing from any of them for many many months.

 

[roy_bates]

as for the peaple that say fpga is the way to go................i still beleive it will be years before its seen on the amiga if at all.(as an 060)

I'm sadly thinking your right. Well, about the .060 part. In a couple of generations old FPGA stock will be powerfull enough to run 000-030 softcores at "real" speeds. When it gets cheap enough (when people no longer want to have them as storage heaters) they have to end up somewhere, and if we are still around, it makes sense to use them to replace hard to find CPU's with cheaper if yet more complex counterparts. I guess only time will tell, but thats what i think. And quite frankly, i don't really see a huge market for 68060 based parts softcore or other-ways, where as a 68030 will keep on selling for as long as there is Amigas without one... You don't really need more when all you do is WHDLoad, and (i might be biased) that seams to be what most people do. Amiga OS4 does enjoy a faster CPU still, but thats a more limited user-group. You guys are more exlusive, while the general public wants the retro experience, to relive some youth.
B!

exactly!,thats why im saying its not a good idea at the moment as most are only interested in gaming.

---------- Post added at 12:07 ---------- Previous post was at 11:31 ----------

It will be years before the price of an FPGA capable of taking a the full 060 design and running at plus 100MHz will be commercially attractive over the (now €80) MC68060RC50

The problem is not only the price of the FPGA's but also the availability of the design (i.e. it is closed source) and the lack of people interested enough to reverse engineer and write HDL Synthesiseable CPU cores.

It's what six years since MiniMig first came out?? Only 3-4 engineers have ever said they've been actively working on 680x0 HDL designs. And only Tobias Gubener's TG68 has been released. Nothing from any of them for many many months.

exactly!,my fear is that it will be too much too late.

as is the never ending case with amiga products take for instants,the x1000 and most hardware for the amiga,by the time it comes out its underpowered and overpriced.

so im probably better off sticking to 68k cpu's,as most of the software is based on this and will be near 100% compatible with it.

anything else as far as im conscerned is an emulation of this.

ppc is/was a good step forward but wasent really implimented well enough for what it was, becuase its risc means it has to have high clock cycles to make it realistic.also the os was a bit.........well lacking,nice to look at but lacking.

i think the problem was it wasent used/accepted as a main cpu to start out with,just a gloryfied fpu for a handfull of programmes and it diddent get an os for years.

as for X86,well theres not much point in it on an amiga in hardware.
we already have winuae.


i think the main problem really is acceptance,most would like the amiga the way it was.the way it felt.the way it looked.the way it sounded.
and for the most part thats the way it stayed.
no matter what we shoehorn into them:lol

 

[keitha]

I remember buying my Apollo 30 in 99/00, think it cost me £160. The 40 was 200 and odds and 06 was about £450-500. In 2000: £500 was about 2 weeks wages! in today's money: that's something like £600-£700. Of course 06 were more common in 2000, than they're in 2012

On a positive note: I was reading on EAB about a Russian guy who built a A600 memory expansion for £60 pounds, and a Greek guy and a guy from Eastern-Europe, who built a A600 010-acc with 4mb memory for something like, for 70 euro's.

I suppose it can be done! I would def be willing to fork-out £200-£250 for a cheap 06 card

 

[Renegrade]

Thoughts on x86 Amigas: Although I loathe and hate X86, the braindead little CPUs in that line are ..very.. fast these days.

However, with a custom motherboard and firmware, one could overcome one of the PC's biggest disadvantage: non-standardized, low-quality firmware and a lack of motherboard resources.

Much of the PC spec is still based on the IBM PC AT™ (Model 5160 I think, please don't sue if I have the number wrong though heh), which was more or less the last common denominator aside from a handful of industry standards that were forced into existence by very very very sharp necessity.

Here's some issues:
- The PC really only has one on-board timer available for use that's guaranteed to be there and NOT used for something else (like dram refresh or speaker output). It got to be such an issue that Intel actually started making on-CPU timers. Putting in a better timing chip would be spiffy!
- The PC has no VSYNC interrupt. This might not be solvable though if one goes with an off-the-shelf "VGA" chip.
- The PC boots in real mode, and the non-EFI bioses live exclusively just to load that boot sector into ram and execute it. MBR partitions are like a nightmare from the 8-bit era.
- PC plug and play is a disaster. This might not be solvable though as third party vendors are involved (that's the pattern here..). DSDT tables are poorly defined. The spec is a nightmare of errors and omissions and is taller than a normal human when printed out and stacked up.
-There's like nine different incompatible ways of specifying what memory a PC has. A PC only has one type of memory. There are bizarre memory holes all over.

If one went whole hog and basically designed a system around an x86 chip, one could probably build an Amiga-like experience. That would require having a GPU custom-designed for it, and all sorts of things though.

Of course, it would be rather cool knowing that your CIAs are handling interrupts from the USB mouse without trashing CPU caches, storing mouse pointer coordinates in a pair of hardware registers and then having that passed to the GPU during the VBLANK interrupt for display, without ever having crossed the CPU.

Eh?

 

[Ivan]

On a positive note: I was reading on EAB about a Russian guy who built a A600 memory expansion for £60 pounds, and a Greek guy and a guy from Eastern-Europe, who built a A600 010-acc with 4mb memory for something like, for 70 euro's.

I suppose it can be done! I would def be willing to fork-out £200-£250 for a cheap 06 card

060 card is not a problem at all, the problem is 060 itself is rare
laboriousness of 060 card is the same as 040 and a little bit more than 030
but no cpu

When we are speaking of 060, we should take into account this is last century. And no problem to repeat design of Bliz 060, but it will be waste of time for developer.
Even ColdFire is more interesting, because it is cheap, has 266 MHz and nobody have done this accelerator.
But CF needs patching kickstart and developing coldfire.library and time spending for these is much more than to repeat 060 design

 

[Bamiga2002]

What we ultimately want here is Natami with an extra PPC-cpu socket in it. That's what I would want anyway